Organisation internationale de normalisation

Process for determining Reference Quality bitstreams (RQx)
At any time, if evidence is presented (i.e. listening test results) that show that an Alternate Encoder (AE) shows an improvement over RQx at the 95% level of significance, then AE becomes the new Reference Quality Encoder (RQ_x+1).

5.2.4Exploration

Stephan Schreiner, FhG, presented

m15875

A proposal for requirements for a audio metadata system within MPEG

Stephan Schreiner Wolfgang Fiesel Matthias Neusinger

The contribution notes that HE-AAC is getting widespread adoption in the video broadcast marketplace. However, relative to some competing audio coding systems, HE-AAC has limited support for metadata and which in turn might limit market adoption. The metadata of greatest interest is information to control the following three factors (the three "classical" D's):

• 
  dialog normalization
  
• 
  dynamic range control
  
• 
  downmix
  

Such Audio metadata helps the receiver to manipulate the received audio signal based on the adjustments performed by the listener.

For example, this metadata might can be used for three major applications:

• 
  Midnight Mode – reduced dynamics
  
• 
  Downmix – to derive stereo or mono from a multichannel program
  
• 
  Clean Audio – boost dialog and suppress ambiance for e.g. hearing impaired listeners
  

The contribution reviewed the limitations of the metadata currently supported by AAC (and hence available to HE-AAC) and proposes a potential requirements for a potential metadata activity:

• 
  Backward compatibility – reuse parameters from existing production infrastructure
  
• 
  Computational complexity – reuse existing coding tools to facilitate user control (e.g. filterbank)
  
• 
  Bitstream overhead – metadata should impose low overhead
  
• 
  User interface – should be simple to use
  
• 
  Embedding in coded representation (bitstream) – retain backward compatibility
  

There was considerable discussion on this topic. An issue was raised as to whether the metadata is agnostic to and independent of the coded representation. Another comment was whether metadata is even needed to provide the user control functionality.